First off, lest I get myself into deep doo-doo, I need to state very clearly that what I'm about to present is exploratory conjecture and nothing like a complete design. Don't go building something using a ruler on your screen and your tongue sticking out of the corner of your mouth, scaling from my sketch, and expect it to be safe. That's not what the following is about, so don't sue me. I've got a lot of thoughts I keep coming back to, and this is one of them. Especially in the case of our pre-WWII cars, with their often emphatically non-structural bodywork, there is a lot to be said for having the safety belts anchored to the seats themselves. You look for suitable anchor points, and you find that there aren't any, and putting them in might compromise a lot of other stuff. So anchor points on the seat make sense, even if it means that anchoring the seats to the frame becomes much more onerous. I'm not going into seats with fore-aft adjustability: OEMs do adjustable seats with built-in belt anchors, but absent any useful info on the structural capabilities of seat tracks I wouldn't want to go there. It turns out that tilting backs are easy, though. Two things have always stopped me in the past: 1. figuring out what kind of deceleration and consequent inertial forces to design for, and 2. designing a backrest which is an extremely stout cantilever rooted in the car's floor, especially if we want it to tilt. But it isn't necessary to work that way. Working on the principle of a chain, where the strength of one link is known to be sufficient it makes no sense to make the other links more than a little bit stronger. And we've got a known link: we know the belt can handle about 2¼ tons, and that they actually do work. A three-point diagonal shoulder belt has saved my own ass more than once, and anything better than that is a bonus. If the seat holds up with a bit more than 2¼ tons tugging on the belt, we're golden. The chain thing is for some reason a highly unorthodox approach. I don't quite understand why. It should be obvious that if one thing works, something which is demonstrably equivalent to it should also work. But for some reason you can't work that way. It is somehow related to the contention that round tubing is always and everywhere stronger than square tubing, when virually every available square tubing is stronger even in its weakest axis than round tubing of the same weight. I suspect that it has to do with square-tube junction details being more prone to stress risers unless they are horribly clunky. But I digress. Nor is it necessary for the seat back to be a hectic vertical cantilever. It would if the belt were anchored to the top of the backrest; and then my idea of using a car door latch to secure the backrest wouldn't fly either, because those are only good for a bit over a ton. But if the belt were to be anchored to the floor behind the seat, the seat back becomes a vertical cantilever truss of which the belt becomes a second tensile chord, the bit holding the latch becomes a very lightly-stressed tensile chord, and the main hoop becomes the compressive chord. The main hoop needs to be able to carry about 4 tons in its own plane, but it turns out that that isn't a hell of a lot in the scheme of things. The 1¼" tube in the sketch should do that: I've gone for a 4-point harness here because it's a step up from a 3-point shoulder belt. That should be fine as long as the cushion-to-backrest angle doesn't exceed about 110°. A 5-point harness would be better at greater angles but they're not really practical for the possible range of ordinary clothing. My wife likes her long skirts etc. The inertia reels allow the backrest to tilt but the belt still to serve as a tensile chord in an impact situation. Strictly speaking a 4-point harness with a Y-belt and a single inertia reel would be possible. I've never used a Y-belt but besides feeling like they might be uncomfortable I understand that they are frowned upon these days safety-wise. Hence two inertia reels per seat. The reels are calibrated to lock at between 0.75g and 1.5g, which seems like a wide range and got me worried that they might not lock at the same time in an emergency. As dual-reel harnesses are common in aircraft setups I assume that they work fine because in emergencies we'd be looking at a lot more deceleration than 1.5g. Now, the dual belts mean that we've got 4½ tons capacity, and I can go crazy and design everything to handle that, but I'm wondering if it's worth it. A single belt works, after all. Of course everything needs to be attached very securely, not to the floor but to the frame. This is likely to involve lateral bracing in the order of about 1½" square, under or above the floor, depending on the situation. And I need to point out that all the forces and sizes I've used here pertain to a specific low-slung seat situation —which I've deliberately not dimensioned in any detail— and that taller seats are likely to be chunkier overall. Last of all: the headrests. I like to have them, even if my wife might suggest that my pig-neckedness renders me immune to whiplash. She certainly isn't immune to whiplash. Headrests in a traditional idiom are tricky, but perhaps it is possible to be discreet, so that they don't look all that out-of-place on Vintage-ish buckets like I've drawn. My problem is that in a very low body they are likely to undermine the tilting action by hitting the headliner. Hence the vague conjecture around headrests which fold forward unless locked in place by the backrest being latched upright, via some more or less simple mechanism.
depending upon the rules of your sanctioning body were you're racing, you may need an anti submarine belt. Sent from my SM-G965U using The H.A.M.B. mobile app
Is the ≤110° angle sufficient to eliminate submarining? I'm not an engineer, and I don't play one on TV, but I wouldn't bet my lunch money on it. The design of the three point means that the torso's forward motion tightens the belt at the hips, giving an anti-submarine effect for a properly worn belt. This isn't true of the 4-point. I'm curious why you think a Y-belt would be uncomfortable. The part the user experiences is virtually identical between the two. If the two reels lock in quick sequence, I doubt you'd notice it. The problem comes when only one locks, causing a twisting motion in the upper torso. I suspect that the construction, calibration and testing regimes for the aircraft parts are far stricter than those for automotive use.
dont mount a harness like that you get downward force on your spine in an accident. the rest of it , dont try and re-invent things that existing race rules can show you how to do safely. follow your favorite racing's engineering rules.
It probably also depends on the cushion angle. The steeper that is, the more inertia will tend to push the occupant down onto the cushion rather than forwards along it. Look at the driving position in the latest F1/sports prototypes/etc.: it's much like an upright bus-driver posture, only tilted back with the feet higher than the H-point, resulting in a very steep "cushion". I actually got the calibration figures from aircraft usage.
Of course an anti-submarine belt can be added, which lives under the seat cushion for ordinary street use.
Most sanctioning bodies' rules want the shoulder belts pulling between horizontal and about 45° down. They emphatically don't want an upward angle — though that is common in aircraft, probably because aircraft seating tends to be more upright. I was very much aware of racing practice in this: but it isn't magic; it isn't beyond the ken of mere mortals, it's actually fairly elementary engineering.
Some engineering, standards and testing https://sfifoundation.com/protectivegearrestraintsnets/ https://www.sfifoundation.com/wp-content/pdfs/specs/Spec_16.1_022614.pdf https://www.sfifoundation.com/wp-content/pdfs/specs/Spec_16.5_122914.pdf https://www.sfifoundation.com/wp-content/pdfs/specs/Spec_16.6_042018.pdf
The Jalopy Journal
